Discussion

To the best of our knowledge, this is the largest study to evaluate the relationship between OS and changes in RAS and BRAF status using ctDNA in patients with mCRC. In this study, the median OS associated with Neo_RAS_ and NeoBRAF WT mCRC was shorter than but not significantly different from that of patients with RAS WT. Other RAS and BRAF mutation cohorts had significantly shorter OS than did patients with RAS WT. Furthermore, patients with Neo_BRAF_ WT had a significantly longer median OS than those with persistent BRAF MT. Moreover, non-RAS gene alterations were more frequent in the acquired RAS MT cohort, especially for genes related to the RAS-RAF-MAPK and PI3K-AKT-mTOR pathways.

The absence of detectable ctDNA after treatment is associated with a longer OS.24,25,26 One of the reasons for this is that the ctDNA VAF correlates with tumor volume,27 and patients with a low tumor volume may have a good treatment response and prognosis. In our study, low ctDNA VAF after initial therapy was indeed associated with better OS in the multivariable analysis. This may reflect relative tumor sensitivity before chemotherapy or a less aggressive, slower-growing phenotype. To better understand whether ctDNA-negative cases explained the lack of statistical difference between the Neo_RAS_ WT and RAS WT cohorts, we compared OS for Neo_RAS_ WT Group B, which included only cases with ctDNA evidence of non-R__AS alterations, to the full RAS WT cohort and found similar results. Therefore, the absence of RAS MT in ctDNA after chemotherapy may be associated with improved survival in mCRC. The tumors of patients in this group may have a similar biology to those of patients with RAS WT, raising the hypothesis that tumors in this group may share biological features with RAS WT mCRC, though whether this translates into anti-EGFR responsiveness remains to be prospectively confirmed. As previously reported, an anti-EGFR antibody was administered to six patients in this cohort and was effective in three patients.22 Several clinical trials are ongoing in this setting,28 and the OS may potentially improve if ongoing clinical trials confirm the efficacy of anti-EGFR antibodies in this setting; however, this remains speculative pending the availability of definitive clinical evidence.

Loss of BRAF V600E after chemotherapy may also contribute to improved survival. Among patients with BRAF MT detected by tumor tissue testing at baseline, 18.4% had no BRAF MT in ctDNA after treatment, a phenomenon known as Neo_BRAF_ WT mCRC.29 Regarding the relationship between post-treatment ctDNA status and treatment prognosis, biomarker analysis in the FIRE 4.5 study showed that treatment response and outcomes were better in patients with decreased BRAF MT VAF after treatment and in those who did not have BRAF MT in post-treatment ctDNA.30 Furthermore, some patients with BRAF MT displayed a good prognosis, in contrast to the traditionally poor prognosis.31 Therefore, the confirmation of BRAF MT status in post-treatment ctDNA may also help estimate prognosis and optimize the treatment strategy in BRAF MT mCRC.

Patients with persistent RAS MT had shorter OS than those with RAS WT. RAS mutations are poor prognostic factors for patients with CRC and poor predictors of response to anti-EGFR antibodies.32,33 In addition, patients with acquired RAS MT had shorter OS than those with RAS WT. The main reason for this is that acquired RAS MT is among the factors causing resistance to anti-EGFR antibodies.34 Moreover, the rate of non-RAS alterations was significantly higher in the acquired RAS MT group than in the other groups. In particular, alterations related to the RAS-RAF-MEK, PI3K-AKT-mTOR, and other pathways such as myelocytomatosis oncogene (MYC) were significantly higher in this group. These are often reported to cause resistance to both the primary and acquired alterations associated with anti-EGFR antibodies.35 In cases where these alterations occur, various pathways are activated, and extremely complex resistance mechanisms are thought to lead to reduced OS. Furthermore, acquired RAS MT also reportedly affects the therapeutic efficacy of re-challenge with anti-EGFR antibodies in later-line treatment.

In a sub-analysis of an anti-EGFR antibody re-challenge study, Cremolini et al. reported that treatment efficacy was lower in patients in whom ctDNA RAS was detected before anti-EGFR antibody re-challenge.36 Furthermore, even when ctDNA was restricted to RAS WT before anti-EGFR antibody re-challenge, such as in the REMMARY and PERSUIT trials, which examined the efficacy and safety of anti-EGFR antibody re-challenge, no cases were reported where anti-EGFR antibodies were effective under conditions where new RAS MT appeared even transiently.37 To improve the outcomes of patients with acquired resistance, using molecular-targeted drugs in combination with chemotherapy and avoiding their use alone is preferable.38 This is likely because, when chemotherapy is not used in combination, acquired resistance can develop more easily, even when treatment lines differ. This is likely a major reason why cytotoxic chemotherapy has been selected in several ongoing clinical trials of molecular-targeted drugs for BRAF, HER2, and KRAS MT mCRC. Another approach is to use MT-specific inhibitors that target acquired resistance MTs; preclinical studies have shown that this has a therapeutic effect.35 However, as few studies have targeted subclonal MTs in clinical settings,39 further research is expected.

Conclusion

Post-treatment circulating tumor RAS and BRAF MT dynamics may serve as useful prognostic indicators, particularly with regard to BRAF MT.

Limitations of the study

Our study has some limitations. First, the retrospective design introduces inherent selection bias and limits causal inference, as data collection and patient selection were not prospectively controlled; furthermore, the validity and reproducibility of diagnosis could be improved in cases with low VAF, particularly those with <1%VAF.40 Second, the cohort exhibited considerable heterogeneity with respect to prior treatment history, tumor characteristics, and ctDNA sampling timing, which may have confounded the prognostic associations observed. In addition, immortal-time bias was present, as patients with poor prognosis were not included in this study; to mitigate this, analyses were restricted to patients who underwent post-treatment ctDNA assessment, though residual bias cannot be excluded.41 Therefore, OS was extended beyond the previously reported period. Third, ctDNA was not measured serially before and after each treatment cycle; thus, whether ctDNA and tissue RAS status matched was not confirmed. The absence of serial paired tissue-plasma sampling further limits mechanistic interpretation and prevents definitive conclusions regarding clonal evolution under therapeutic pressure. Finally, the lack of independent external validation represents a critical limitation; these findings should therefore be considered exploratory and require prospective confirmation in larger, independent cohorts before recommending clinical application. Despite these limitations, our results suggest that confirming gene alterations in ctDNA after treatment may be useful for estimating prognosis and optimizing treatment selection in mCRC.